查看文章

rowth in the number of fuzzy logic applications has fueled the need for computing systems that efficiently process fuzzy inferences.(For pointers to literature on fuzzy logic." see the box.) Although all-software solutions exist, hardware structure often must support systems to achieve adequate processing speeds for applications from real-time controllers to complex expert systems. Our inference processor design uses a core based on an analog-numerical approach that combines inherent advantages of both analog and digital implementations. Its archi-tectural model features

• optimization of the number of rules processed for each inference by means of an on-line preprocessing technique and

• parallel inference execution.